Download SDT tobacco cessation

Health Psychology
2006, Vol. 25, No. 1, 91–101
Copyright 2006 by the American Psychological Association
0278-6133/06/$12.00 DOI: 10.1037/0278-6133.25.1.91
Testing a Self-Determination Theory Intervention for Motivating Tobacco
Cessation: Supporting Autonomy and Competence in a Clinical Trial
Geoffrey C. Williams, Holly A. McGregor, and
Daryl Sharp
Chantal Levesque
Southwest Missouri State University
University of Rochester
Ruth W. Kouides
Richard M. Ryan and Edward L. Deci
Rochester General Hospital
University of Rochester
A longitudinal randomized trial tested the self-determination theory (SDT) intervention and process
model of health behavior change for tobacco cessation (N ⫽ 1006). Adult smokers were recruited for a
study of smokers’ health and were assigned to intensive treatment or community care. Participants were
relatively poor and undereducated. Intervention patients perceived greater autonomy support and reported
greater autonomous and competence motivations than did control patients. They also reported greater
medication use and significantly greater abstinence. Structural equation modeling analyses confirmed the
SDT process model in which perceived autonomy support led to increases in autonomous and competence motivations, which in turn led to greater cessation. The causal role of autonomy support in the
internalization of autonomous motivation, perceived competence, and smoking cessation was supported.
Keywords: tobacco dependence treatment, autonomous motivation, perceived competence, selfdetermination theory, adherence
change among the subset of patients with elevated cholesterol
(Williams, Minicucci, et al., 2002). This article is limited to the
tobacco treatment outcomes.
Research on SDT (Deci & Ryan, 1985; Sheldon, Williams, &
Joiner, 2003) has indicated that autonomous and competence motivations are associated with change in tobacco use for adolescents
(Williams, Cox, Kouides, & Deci, 1999) and long-term tobacco
abstinence for adults (Curry, Wagner, & Grathaus, 1991; Williams,
Gagné, Ryan, & Deci, 2002). However, no randomized trial has
provided an intervention that has increased patients’ perceptions of
autonomy support, autonomous motivation, and the intended behavioral outcome. Central to SDT are the concepts of autonomous
motivation and perceived competence. People are autonomously
motivated when they experience volition and choice while behaving. When stopping smoking, patients would be autonomous if
they freely chose to stop because they were personally committed
to improving their health and the quality and length of their life.
People are autonomous in using medications if they freely choose
to take a medication they believe may help them reach a valued
health goal. People perceive themselves to be competent when
they feel able to attain important health outcomes. The construct of
perceived competence is similar to that of self-efficacy (Bandura,
1997). We used the perceived competence scale because it was
developed with the other measures used in the SDT process model
and it is short (four items). Studies indicate that when people are
more autonomously motivated they feel more competent to attain
relevant health outcomes (Williams, Freedman, & Deci, 1998;
Williams, McGregor, Zeldman, Freedman, & Deci, 2004). This is
consistent with SDT because autonomy concerns the experience of
initiating behaviors that prompt people to feel more competent in
Results from the Lung Health Study (Murray, Connett, Rand,
Pan, & Anthonisen, 2002) indicate that 21% of smokers who
received intensive tobacco-dependence treatment had maintained
abstinence 11 years later compared with 6% who received community care. However, little is known about what factors predict
abstinence. The Public Health Service’s tobacco-dependence
guidelines (Fiore et al., 2000) suggest that supports for autonomy
and self-efficacy may be important, but the guidelines offer little
empirical support for these process variables. In 1999, the National
Institutes of Health (NIH) funded 15 studies of behavior change,
referred to as the Behavior Change Consortium (BCC), to link
theories of behavior change with important health outcomes (Ory,
Jordan, & Bazzarre, 2002). The current study, which is part of the
BCC, was a randomized trial conducted to test a self-determination
theory (SDT) intervention for tobacco cessation and for dietary
Geoffrey C. Williams, Department of Medicine and Department of
Clinical and Social Science in Psychology, University of Rochester; Holly
A. McGregor, Daryl Sharp, Richard M. Ryan, and Edward L. Deci,
Department of Clinical and Social Science in Psychology, University of
Rochester; Chantal Levesque, Department of Psychology, Southwest Missouri State University; Ruth W. Kouides, Department of Medicine, Rochester General Hospital, Rochester, New York.
This research was supported by Grant 1-R01-MH5954, cofunded by the
National Institute of Mental Health and the National Cancer Institute, and
by National Center for Research Resources Grant 5-M01-RR00044 to the
University of Rochester General Clinical Research Center.
Correspondence concerning this article should be addressed to Geoffrey
C. Williams, Department of Clinical and Social Science in Psychology,
University of Rochester, Rochester, NY 14627. E-mail: geoffrey_williams@
urmc.rochester.edu
91
92
WILLIAMS ET AL.
attaining outcomes. SDT predicts that people will be most effective in stopping smoking when they feel autonomous and competent for doing so.
According to SDT, when practitioners are autonomy supportive,
patients will be more likely to become autonomously motivated
and to perceive themselves as competent for cessation. Autonomy
support refers to practitioners’ eliciting and acknowledging patients’ perspectives, supporting their initiatives, offering choice
about treatment options, and providing relevant information, while
minimizing pressure and control. Studies have shown that autonomy support by health care practitioners affects patient motivation
and health behaviors, including smoking abstinence (Williams,
Cox, et al., 1999), weight loss (Williams, Grow, Freedman, Ryan,
& Deci, 1996), and medication adherence (Williams, Rodin, Ryan,
Grolnick, & Deci, 1998). The present trial examined whether an
intervention focused on practitioners’ bringing patients to a place
of informed choice about whether or not to smoke, and leaving the
decision to the patients, would lead to greater perceptions of
autonomy (i.e., volition) and competence, and in turn to greater
tobacco abstinence.
The relations among autonomy support, autonomous motivation, and perceived competence as predictors of health outcomes
form the basis of the SDT process model of behavior change.
Theoretically, the SDT process model represents a process of
change called internalization, through which people take in and
integrate the motivations and competencies for changing a particular behavior or goal. Internalization is expected to be the active
process of change in all human beings no matter what treatments
(or nontreatments) they encounter. We hypothesized that the intervention would be more likely to facilitate patients’ internalizing
autonomous and competence motivations to stop smoking because
it would enhance the level of autonomy support experienced by
intervention patients as compared with community care patients.
However, we also predicted that, when people experienced autonomy support, their autonomous motivation and perceived competence would increase accordingly regardless of which condition
they are in because internalization is theorized to be a proactive
change process for all people. Thus, the relations among the
primary motivation variables are expected to be comparable regardless of the level of any of the variables. For example, the
magnitude of the relation between autonomous motivation and
perceived competence is expected to be stable regardless of level
of autonomous motivation experienced. Further, it is expected that,
as autonomy support, autonomous motivation, and perceived competence increased as a system, behavior change would be more
likely to occur.
On the basis of this, we hypothesized that the intensive intervention condition relative to the community care condition would
result in (a) greater perceived autonomy support, (b) greater internalization of autonomous motivation for stopping smoking and for
taking cessation medications, (c) greater development of perceived
competence for quitting, (d) greater use of medications for stopping smoking, and (e) greater abstinence rates at 6 months. In
addition, we hypothesized that (f) the predictive relations among
the motivation variables of the SDT process model would be
equivalent across the two conditions and that (g) changes in the
network of motivation variables would predict both greater use of
medications and greater abstinence.
Although SDT has been developed separately from social learning theory (SLT; Bandura, 1997) and stages of change (SOC;
Prochaska, DiClemente, & Norcross, 1992), there are similarities
among some of the constructs used in these theories (e.g., selfefficacy is similar to perceived competence). However, because
change in SDT is hypothesized to occur as a function of internalization of autonomous and competence motivations, and change in
SLT and SOC is hypothesized to occur through other mechanisms
(e.g., increasing outcome expectancies), interventions based on the
three theories tend to be constructed differently to motivate
change. SDT assumes that internalization is the process of change
for all humans, and that humans are inherently motivated to
internalize the regulation of behaviors that allow them to grow and
be healthy. Patients’ perceptions of autonomy and competence are
expected to increase when the patients are supported in making an
informed choice for themselves about whether or not to continue a
behavior or to start taking a medication. Therefore, the intervention designed for this study establishes an autonomy-supportive
climate in the same way for all smokers, no matter their stage of
change or their expectancies of efficacy.
Method
Study Design and Conditions
The study design, recruitment, and detailed description of the intervention were reported elsewhere (Williams, Minicucci, et al., 2002). In brief,
people who smoked five or more cigarettes per day, were 18 years of age
or older, read and spoke English, had no history of a psychotic illness
(depression and anxiety were allowed), had a life expectancy of 18 months,
and planned to live in the greater Rochester, NY, area for 18 months were
recruited through newspaper ads and signs in physician offices to participate in a study about “smokers’ health.” Intention to quit was not required,
as we wanted to work with a general population of smokers, and 52% of
the participants indicated that they were not ready to try quitting when they
entered the study, before contact with the counselors. The participants were
relatively poor, with average incomes more than $10,000 below that of the
average for the county. They were also undereducated, with 11% of the
sample over 24 years of age having graduated from college, compared with
33% for the county.
Parking passes were given to participants, and they were paid $30 after
completing the 6-month questionnaire. (The honoraria were prorated if
participants withdrew from the study before completing it.) Participants
were stratified according to their cholesterol levels (National Institutes of
Health, National Heart, Lung, and Blood Institute, 1997) and were randomly assigned to either community care (30% of the sample) or to the
intensive intervention (70% of the sample). This ratio was used because
intensive interventions were expected to have greater effect on cessation
than community care, and we would minimize the number in community
care while still allowing for a test of mediation. All community care
patients will be offered intensive treatment once they finish the study.
The community’s accepted standard of care was selected as the comparison group because it would not be ethical to intentionally undermine
patient autonomy with a controlling condition. Thus, the main questions
addressed in this study are what happens to patients’ autonomous motivation and perceived competence and to their subsequent health behaviors
(taking medications and stopping smoking) if they are provided with
greater autonomy support than is provided in the community.
Community care condition. Participants randomized to community
care received the Public Health Service (PHS) booklet You Can Stop
Smoking (PHS, 2000), a photocopy of the results of their cholesterol tests,
and a list of active smoking cessation programs (phone numbers and costs)
in the Rochester area. They were advised to make an appointment with
MOTIVATING TOBACCO CESSATION
their own physician to discuss their cholesterol results and smoking, and
they were encouraged to enroll in a smoking cessation program.
Intensive intervention condition. Participants in the intervention condition were given the same recommendations and materials given to those
in community care and were asked to meet with study counselors four
times in the subsequent 6 months. The first meeting was in person. The
subsequent meetings were also encouraged to be in person, but phone
follow-up was allowed if the participants so chose. Additional in-person or
phone contacts with a counselor were allowed if initiated by the patients,
and patients were told that they could schedule a meeting with a study
physician for a medication consultation if they chose to do so. After
completing baseline questionnaires (approximately 50 min), the participants had their first meeting with a health counselor (50 min).
After the first visit, the patient met with the counselor for 20-min
follow-up visits. Counselors had been trained to support the participants in
making a clear and autonomous decision about whether or not to make a
quit attempt. The participants’ smoking history and attitudes were explored, including why they liked smoking, their past quit attempts, and
what health risks they perceived from smoking. Counselors asked briefly
about participants’ life strivings and about how smoking either helped or
hindered their attainment. Participants’ 10-year risk for developing cardiovascular disease was reviewed with them (Grundy, Pasternak, Greenland,
Smith, & Fuster, 1999), and they were informed that quitting smoking
would cut that risk by 50% within 12 months (U.S. Department of Health
and Human Services, 1990).
After summarizing and acknowledging any reactions from the patients,
counselors asked whether or not the patients wanted to stop smoking. If
they were not ready to stop, they were asked to return in 1 to 2 months to
discuss smoking further. If they said they wanted to stop smoking, counselors moved on to support their perceived competence for quitting by
establishing a cessation plan according to Public Health Service (PHS)
guidelines (Fiore et al., 2000). Specifically, counselors discussed setting a
quit date in the next 30 days, informed the patients about the expected time
course of withdrawal, and discussed the medications recommended to
reduce withdrawal symptoms. Follow-up was planned for the 1st week
after cessation. Patients were also asked if they wanted to discuss their
smoking and medication use with a study physician or with their own
physician.
In all parts of the interview, counselors focused on understanding the
patients’ perspectives and emotional responses. Follow-up visits for those
who had not wanted to quit involved counselors’ reviewing the participants’ values and initiating a discussion about any further thoughts on
whether to try quitting. At any time the patients said they wanted to quit,
counselors pursued supporting competence and making a quit plan.
Follow-up visits for those who wanted to quit focused on reviewing and
acknowledging any quit attempts, reframing failed attempts as short successes and asking if they wanted to quit again, asking about any negative
medication effects, and reviewing standard relapse prevention according to
the PHS guidelines (Fiore et al., 2000).
Procedure and Time Line
Baseline assessments. At baseline, all participants completed questionnaires assessing demographic information, medical history, smoking history, and intention to quit smoking in the next 30 days. They also had their
blood pressure measured. In addition, participants completed the Fagerstrom Addiction Severity Scale (FAS; Fagerstrom & Schneider, 1989), the
Treatment Self-Regulation Questionnaire (TSRQ) for autonomous motivation (Williams, Gagné, et al., 2002), and the Perceived Competence Scale
for Cessation (PCSC; Williams, Gagné, et al., 2002).
1-month assessment. One month after baseline, participants were
mailed a follow-up questionnaire packet. Three attempts were made to
phone participants who did not return questionnaires in 2 weeks, and a
second questionnaire was mailed if needed. The 1-month packet assessed
smoking status and intention to quit smoking, as well as the TSRQ and the
93
PCSC. The packet also included the Health Care Climate Questionnaire
(HCCQ; Williams, Grow, et al., 1996) to assess patients’ perceptions of the
health care practitioner’s autonomy support regarding tobacco use.
6-month outcomes. Six months after baseline, all participants were
mailed a follow-up questionnaire that asked about quit attempts, smoking
status, point prevalence, and the use of medications. Participants who
reported not smoking via the point prevalence question were instructed to
call the study to arrange for a validation test.
Measures
Smoking status, cessation, and medication use. Smoking status and
cessation were assessed as follows. The primary measure was biochemically validated 6-month 7-day point prevalence (7-day PP), which is a
dichotomous measure. Patients were asked to respond either “yes” or “no”
to the following question: “Have you smoked a cigarette, even a puff, in the
past 7 days?” For those participants who answered “no,” serum cotinine
was analyzed to validate the point prevalence report of cessation (Pojer et
al., 1984). As recommended (Hughes et al., 2003), we report both the
validated and unvalidated 7-day PP. Only those participants who responded
“no” to the 7-day PP question and were validated as not smoking via serum
cotinine were treated as not smoking.1 Participants were also asked to
report the longest number of days they had not smoked since they began
the study and the number of days since their last cigarette. A latent
cessation variable for use in the structural equation modeling (SEM)
analyses was constructed using validated 7-day PP, longest number of days
not smoking, and number of days since last cigarette as indicators. The
three indicator variables have been shown to relate to long-term abstinence
in previous reports (Dale et al., 2001; Farkas et al., 1996; Williams, Gagné,
et al., 2002) and loaded significantly on the 6-month latent cessation
variable in this study (␤s ⫽ .90, .81, and .85, ps ⬍ .01, respectively).
In addition to assessing “cessation,” we measured 6-month “prolonged
abstinence” (6-month PA; Hughes et al., 2003), because it has been found
to be more stable (Pierce & Gilpin, 2003). Specifically, whereas 6-month
cessation was assessed with 7-day PP 6 months after patients began the
study, 6-month PA was assessed 6 months after the patients’ reported quit
dates (Hughes et al., 2003). Finally, we asked the number of serious quit
attempts (lasting more than 24 hr) the patients had made since beginning
the study.
Participants were also asked if they had used any of the following
medications to aid in cessation: nicotine patch, nicotine gum, nicotine nasal
spray, nicotine inhaler, or sustained release bupropion. For each medication
used, participants reported the type, the dose, and the number of days it was
used. From this, two variables were created. First, the numbers of days
each medication was used were summed to obtain the number of medication days. The second variable was the number of medications used by the
participant. These two variables were significant indicators of the latent
medication variable in the SEM analysis (␤s ⫽ .50 and .50, ps ⬍ .01).
TSRQ. Ryan and Connell (1989) developed the TSRQ to assess autonomous motivation, and it has been described (Williams, Minicucci, et
al., 2002) and validated elsewhere (Williams, Gagné, et al., 2002) Autonomous motivation for cessation and autonomous motivation for taking
cessation medication (nicotine replacement or sustained release bupropion)
were each assessed using six items at baseline and at 1 month. Participants
responded to each item on a 1 (strongly disagree) to 7 (strongly agree)
scale (example item: “The reason I would stop smoking permanently or
1
Across our intention-to-treat sample of 1,006 participants, 118 answered “no” to the 6-month assessment of the 7-day PP question. Of these
118 participants, 96 (81%) were validated as not smoking by serum
cotinine. Of the 22 participants who reported they were not smoking, but
were not validated as such, 17 were in the intensive intervention and 5 were
in community care. Each of these 22 participants was considered as
smoking in the final analysis.
WILLIAMS ET AL.
94
continue not smoking is because I feel that I want to take responsibility for
my own health”). Items reflecting autonomous reasons for stopping smoking exhibited good internal consistency at baseline (␣ ⫽ .86) and at 1
month (␣ ⫽ .89), as did those for using medications at baseline (␣ ⫽ .87)
and at 1 month (␣ ⫽ .91). A factor analysis was conducted at both baseline
and 1 month in which all 12 items from both scales were included. At both
time points, autonomous motivation for taking medications and autonomous motivation for cessation loaded cleanly on their respective factors,
yielded eigenvalues greater than 1, and had no cross-loadings greater than
.25.
The means of the six items were calculated for each scale at each time,
yielding baseline and 1-month scores for autonomous motivation to quit,
and baseline and 1-month scores for autonomous motivation for taking
medication. The two scale scores at baseline and the two scores at 1 month
were significant indicators of the latent variables of autonomous motivation at baseline and autonomous motivation at 1 month, respectively (␤s ⫽
.50 and .66, ps ⬍ .01, at baseline, and .52 and .70, ps ⬍ .01, at 1 month).
PCSC. Four items assess the degree to which patients feel able to stop
smoking successfully. Respondents indicated their agreement with each
item on a 1 (strongly disagree) to 7 (strongly agree) scale (example item:
“I feel confident in my ability to stop smoking permanently”). The four
items exhibited good internal consistency at baseline (␣ ⫽ .91) and at 1
month (␣ ⫽ .93). The four items were averaged at each time point to form
baseline and 1-month perceived competence indices for the analyses of
variance (ANOVAs). The four items at each time point were used separately as indicators of the latent variable for perceived competence at
baseline and 1 month in the SEM analyses. Each of the indicators loaded
significantly on its respective latent variable (␤s ⫽ .87, .87, .82, and .89,
at baseline, and .91, .90, .86, and .94 at 1 month).
HCCQ ( Williams, Gagné, et al., 2002; Williams, Grow, et al., 1996 ).
The HCCQ assesses patients’ perceptions of the degree to which their
providers were autonomy supportive (vs. controlling) in consulting with
them regarding their smoking. Patients responded to 15 items on a 1
(strongly disagree) to 7 (strongly agree) scale (example item: “I feel that
my counselor has provided me with choices and options about my smoking”). The mean of the 15 items in the present sample was calculated to
create the HCCQ index used in ANOVAs. The scale had good internal
consistency (␣ ⫽ .94). The latent variable for autonomy support was
created by averaging three mutually exclusive sets of 5 items each from the
HCCQ, and each of these was used as an indicator of the latent autonomysupport variable (␤s ⫽ .87, .95, and .88, ps ⬍ .01).
4.31), t(1002) ⫽ 2.30, p ⬍ .05. For non-White minorities, there
was no difference between those who dropped out of the study and
those who stayed in the study on either perceived competence for
quitting or autonomous motivation to use medications (all ps ⬎
.35). However, minorities who dropped out reported higher, yet
nonsignificant, autonomous motivation for quitting smoking at
baseline than those who stayed in (6.63 vs. 6.36, Cohen’s d ⫽ .37),
t(179) ⫽ 1.71, p ⫽ .09.
Analytic Overview
Two types of analyses were conducted with these data. First, we
conducted an “intention-to-treat” analysis, which included all
1,006 participants. Missing data regarding the motivation variables
were replaced by a participant’s last known report or, if necessary,
by mean replacement. If smoking status or medication-taking
information was unavailable at the 6-month follow-up, the participant was considered smoking or not taking medication,
respectively.
Second, the SDT process model was tested using an “as treated”
subsample of our 1,006 participants. For these 866 participants
who were active at 1 month, we required those in the intervention
to have completed the intervention (at least four contacts) to be
included. If smoking status or information regarding medication
use for any of these 866 was not available at the 6-month followup, participants were considered to be smoking or not using
medications. Multigroup SEM analyses were used to test the SDT
process model on this subsample, as it offered the greatest degree
of data actually reported by participants who had completed the
full intervention. Stated differently, inclusion of people who did
not experience the full intervention provides a poor test of the
process model concerning changes in autonomy and competence,
because they would not have experienced the full intervention.
Further, “pulling forward” their data would create a bias toward
supporting the hypotheses, because using their data would increase
power without adding variance in motivational change.
Preliminary Analyses
Results
Recruitment and Retention
Of the 2,681 individuals screened for eligibility, 1,006 were
randomized to condition. Of the 1,006 randomized into the study
(292 community care, 714 intensive intervention), 866 were active
at 1 month (250 community care, 616 intensive intervention), and
703 were active at 6 months (207 community care, 496 intensive
intervention). During the 6 months following randomization, 78
individuals withdrew from the study, 6 died (no deaths were
related to the study), and 219 were lost to follow-up at 6 months
(see Figure 1).
A comparison of participants active at 1 month with those who
dropped out (n ⫽ 140) indicated that those who dropped out of the
study were less likely to be Caucasian (73.6% vs. 83.2%), ␹2(1) ⫽
7.52, p ⬍ .01; were more likely to have said they were ready to
quit in the next 30 days (55.4% vs. 45.6%), ␹2(1) ⫽ 4.58, p ⬍ .05;
reported higher autonomous motivation for quitting smoking at
baseline (6.26 vs. 6.05), t(1002) ⫽ 2.16, p ⬍ .05; and reported
higher perceived competence for quitting at baseline (4.65 vs.
The 1,006 participants included in the intention-to-treat analyses
had a mean age of 45.5 years, and 63.9% were female. Randomization was effective, as t tests and chi-square analyses showed that
the groups did not differ significantly on key demographic variables (see Table 1) or on baseline motivation variables (see Table
2). Those randomized to the intervention (N ⫽ 714) had an average
of 4.42 visits with a counselor (69% had 4 or more visits).
Fifty-one percent of the visits were in person. Of the 714 participants in the intervention, 323 chose to see a study doctor (45%).
Participants choosing to see a study doctor had an average of 1.3
visits with the doctor, and 33% of those visits were in person rather
than via phone.
Intention-to-Treat Analyses
Effect of the intervention on smoking outcome variables. As
shown in Table 3, chi-square analyses confirmed the central hypothesis that the abstinence outcome of 6-month validated 7-day
PP was significantly higher in the intensive treatment condition
than in community care (11.8% vs. 4.1%), ␹2(1) ⫽ 14.07, p ⬍
MOTIVATING TOBACCO CESSATION
Figure 1.
95
CONSORT recruitment and retention of participants.
.001; number needed to treat (NNT)2 ⫽ 13.1. Similarly, 6-month
PA differed by condition (11.2% vs. 3.8%), ␹2(1) ⫽ 13.93, p ⬍
.001; NNT ⫽ 13.5. The 6-month nonvalidated 7-day PP was also
significantly higher in the intervention than community control
(14.1% vs. 5.8%), ␹2(1) ⫽ 13.87, p ⬍ .001; NNT ⫽ 12.0. Furthermore, t tests confirmed that participants in the intensive intervention condition had significantly more days since their last
cigarette (20.1 vs. 5.7 days), t(1004) ⫽ 4.15, p ⬍ .001, and had
more continuous days not smoking in the first 6 months of the
study (25.1 vs. 9.4 days), t(1004) ⫽ 4.99, p ⬍ .001. The 1-month
7-day PP was also significantly increased by the intervention
(19.5% vs. 3.8%), ␹2(1) ⫽ 33.087, p ⬍ .001. Thus, in terms of
smoking cessation, the intervention had significant effects.
Effect of intervention on medication taking. Chi-square analyses showed that a higher percentage of participants in the inten-
sive intervention reported making a serious quit attempt (49.7% vs.
39.0%), ␹2(1) ⫽ 9.47, p ⬍ .005, and taking medications (30.8%
vs. 15.8%), ␹2(1) ⫽ 24.16, p ⬍ .001. A t test confirmed that these
participants reported taking the medications for more days (29.1
vs. 9.2), t(1004) ⫽ 4.75, p ⬍ .001.
Effect of the intervention on motivation variables. As shown in
Table 2, t tests confirmed that patients in the intensive treatment
group perceived greater autonomy support than those in community care, t(1004) ⫽ 6.19, p ⬍ .001. Analysis of covariance
(ANCOVA) revealed that intensive-treatment patients internalized
2
NNT represents the number of participants that need to be treated in the
intensive intervention that results in 1 additional participant to stop smoking for 6 months.
WILLIAMS ET AL.
96
Table 1
Baseline Characteristics for Both the Intention-to-Treat Sample and the As-Treated Sample by
Treatment Group
Characteristic
Community care group
(n ⫽ 292)
Intensive treatment group
(n ⫽ 714)
p
Sex (% female)
Age (M)
SES (1–9)
Martial status (% married or living as married)
Ethnicity (% White)
Cigarettes per day (M)
Fagerstrom AS (M)
66.8
44.8
4.34
46.6
80.8
20.9
4.97
62.7
45.5
4.36
47.1
81.9
20.3
5.02
.22
.93
.93
.53
.87
.43
.74
Note.
SES ⫽ socioeconomic status; AS ⫽ addiction severity.
more autonomous motivation for taking medications, F(1, 988) ⫽
9.42, p ⬍ .01, and more competence motivation, F(1, 1001) ⫽
30.38, p ⬍ .001, from baseline to 1 month than did those in
community care. However, intensive-treatment patients did not
internalize significantly more autonomous motivation for stopping
smoking, F(1, 1001) ⫽ 1.42, p ⫽ .23. A subgroup analysis
revealed that autonomous motivation for cessation was significantly increased for intervention patients who elected to see a
study physician compared with those in community care, F(1,
615) ⫽ 6.25, p ⫽ .01.
Effect of the motivation variables on smoking outcomes. Correlations of the motivation variables with the smoking outcome
variables are presented in Table 4. Logistic and multiple regressions were used to test the mediational effect of the block of
change in motivation variables on the relation between the intervention and the smoking outcomes. Specifically, each smoking
outcome was regressed on intervention condition and baseline
motivation variables in Step one, and then on the 1-month motivation variables in Step two. Mediation is considered to be evident
when the effect of the intervention on the outcome, controlling for
the change in motivation variables from baseline to 1 month, drops
either by a significant degree or to nonsignificance (Baron &
Kenny, 1986).
Using the Wald criterion for z in logistic regression, the intervention condition significantly predicted both 6-month validated
7-day PP and 6-month PA controlling for the baseline motivation
variables, z(1) ⫽ 12.68, p ⬍ .001, OR ⫽ 3.10 (95% confidence
interval [CI]: 1.66, 5.76) and z(1) ⫽ 12.48, p ⬍ .001, OR ⫽ 3.21
(95% CI: 1.68, 6.12). The addition of the 1-month motivation
variables to the equation yielded a significant effect for the block
in both the 6-month 7-day PP, ␹2(3) ⫽ 31.58, p ⬍ .001, and
6-month PA outcomes, ␹2(3) ⫽ 34.40, p ⬍ .001. Intervention
condition at the second step did not fall to nonsignificance, but its
effect was reduced from z(1) ⫽ 12.68, p ⬍ .001, to z(1) ⫽ 7.26,
p ⫽ .01, for 6-month 7-day PP, and reduced from z(1) ⫽ 12.48,
p ⬍ .001, to z(1) ⫽ 6.89, p ⫽ .01, for 6-month PA, indicating
partial mediation.
Ancillary Analyses
These analyses were fully replicated for the 527 participants
who reported at baseline that they did not want to quit smoking,
with the exception that the motivation variables as a block fully
mediated the relation between the intervention and each of the
abstinence outcomes. As before, the intervention group perceived
greater autonomy support (6.21 vs. 5.61, p ⬍ .001), had greater
change in autonomous reasons for using medications (5.19 vs.
4.91, p ⬍ .05), perceived competence (4.24 vs. 3.59, p ⬍ .001),
had greater days on medications (26.75 vs. 3.75, p ⬍ .001),
validated 7-day PP (9.8% vs. 3.7%, p ⬍ .02), and 6-month PA
(9.0% vs. 3.1%, p ⬍ .05). In addition, across those who did not
want to quit smoking, autonomous motivation for cessation was
significantly increased for intervention patients who elected to see
a study physician compared with those in the community care
Table 2
Baseline and 1-Month Mean Differences of Motivation Variables Between Community Care and Intensive Intervention Conditions
Variable
Baseline variables
Intention to quit in 30 days (%)
Autonomous motivation for quitting
Autonomous motivations for taking medication
Perceived competence
1-month variables
Autonomy support
Autonomous reasons for quitting at 1 month
Autonomous reasons for taking medication at 1 month
Competence motivation at 1 month
Community care
Intensive intervention
p
␣
44.3
5.99
5.21
4.30
48.1
6.11
5.30
4.38
.37
.12
.40
.50
—
.85
.87
.90
5.66
6.16
5.21
4.25
6.26
6.22
5.46
4.74
⬍.001
.23
.002
⬍.001
.95
.88
.90
.92
Note. Mean values reported for autonomous reasons for quitting, autonomous reasons for taking medication, and competence motivation are adjusted for
each variable’s baseline covariate.
MOTIVATING TOBACCO CESSATION
97
Table 3
Smoking Cessation Outcomes and Cessation Medication Use by Treatment Group for Both Intention-to-Treat and As-Treated Samples
Outcome
Community
care
Intensive
intervention
Odds
ratio
Confidence
interval
p
Used medication
Days on medications
Validated 6-month PP
6-month PA
15.8%
5.67
4.1%
3.8%
30.8%
20.05
11.8%
11.2%
2.38
—
3.11
3.22
1.67, 3.39
—
1.67, 5.79
1.69, 6.15
.001
.001
.001
.001
Note. PP ⫽ point prevalence; PA ⫽ prolonged abstinence.
group, F(1, 284) ⫽ 6.75, p ⬍ .01. Mediational analyses showed
the addition of the block of 1-month motivation variables predicted
significant variance in validated 7-day PP, ␹2(3) ⫽ 17.88, p ⬍
.001, and 6-month PA, ␹2(3) ⫽ 18.19, p ⬍ .001. Of importance,
the addition of the 1-month variables reduced the effect of the
intervention to nonsignificant (from z ⫽ 5.23, p ⫽ .02, to z ⫽ 2.13,
p ⫽ .15, for validated 7-day PP, and from z ⫽ 5.29, p ⫽ .02, to z ⫽
2.15, p ⫽ .14, for 6-month PA). Thus, the SDT motivation variables fully mediated the intervention effect on the abstinence
outcomes for those who did not want to quit.
On the basis of the findings of the intention-to-treat analyses,
SEM was used to confirm the relations among the SDT mediators.
This SEM analysis examined the relations among the process
model variables for those patients who reported completing four
contacts in 6 months for the intervention group and those in the
community control group who were active at 4 weeks into the trial.
The 6-month 7-day PP between-groups difference for the astreated analysis was 13.6% versus 4.8%, ␹2(1) ⫽ 14.09, p ⬍ .01.
indicator of the latent 6-month cessation variable, a tetrachoric
correlation matrix was used as input for the estimation of the SEM
analyses (see Bollen, 1989, pp. 440 – 445, and Kline, 1998, pp.
237–238). In addition to the evaluation of model fit, standardized
parameter estimates were evaluated to determine whether the hypothesized relationships were supported by the data.
SDT model equivalence across intervention conditions. Multigroup SEM analyses were performed to examine whether the
SDT model differed across the two intervention conditions. Specifically, these multigroup SEM analyses tested whether the relations between the observed indicators and their respective latent
variables (the measurement model) and also the hypothesized
interrelations among the latent variables (the structural model)
were variant versus invariant across the community care and
intervention groups (Byrne, 2001). Preliminary analyses tested the
SDT model in each group independently and determined that the
fits of Model A (community care) and Model B (intensive intervention) were adequate (Byrne, 2001), and thus we proceeded with
the analysis (see Table 5). The SDT model was then tested simultaneously across the two intervention groups, and the factor loadings, variances, covariances, structural paths, and latent means
were allowed to vary freely (Model 0). The fit of Model 0 was
adequate (see Table 5). Model 0 thus represents the best possible
fit of the data to the model.
Model invariance is determined by comparing the baseline
model (Model 0) to successively restrictive models (ones in which
factor loadings, covariance, or structural paths are constrained to
be equal across the groups). A model whose fit is not significantly
different from that of Model 0 (assessed as difference in the
magnitude of chi-square) is deemed invariant or equal to Model 0,
Multigroup Analysis of the SDT Process Model
Model estimation and fit criteria. Structural equation models
of the data were conducted using AMOS 5 (Arbuckle, 2003). The
maximum-likelihood method was used to estimate parameters, and
goodness-of-fit was assessed by examining the following indices:
the chi-square statistic, the incremental fit index (IFI), the comparative fit index (CFI), and the root-mean-square error of approximation (RMSEA). A model is considered to have adequate fit
when the IFI and CFI have values that exceed .90 and when the
RMSEA is less than .08 (Bollen, 1989; Byrne, 2001; Kline, 1998).
Because of the inclusion of the dichotomous point prevalence as an
Table 4
Intercorrelations Between Motivation Variables and Smoking Outcome Variables
Variable
1.
2.
3.
4.
5.
6.
7.
8.
9.
†
Baseline autonomous motivation for cessation
Baseline autonomous motivation for taking medication
Baseline perceived competence
1-month autonomy support
1-month autonomous motivation for cessation
1-month autonomous motivation for taking medication
1-month perceived competence
Longest time not smoking
Days since last cigarette
p ⬍ .10.
* p ⬍ .05.
** p ⬍ .01.
1
2
3
4
5
6
7
8
9
—
.46**
.40**
.19**
.76**
.40**
.38**
.08*
.07*
—
.19**
.13**
.39**
.68**
.20**
.06†
.05
—
.13**
.34**
.12**
.66**
.08*
.09**
—
.23**
.25**
.24**
.12**
.11**
—
.49**
.45**
.11**
.09**
—
.23**
.10**
.08*
—
.28**
.19**
—
.71**
—
WILLIAMS ET AL.
98
Table 5
Goodness-of-Fit Indices From the Multigroup Analysis
Model
Model
Model
Model
Model
Model
Model
Model
Model
Model
Model
Model
Model
A (usual care)
B (intervention)
0 (both groups)
1 (factor loadings constrained)
2 (BL covariance constrained)
3 (all paths constrained to be equal)
4 (BL to 1M motivation paths constrained)
5 (AS to 1M motivation paths constrained)
6 (autonomy 1M to competence 1M constrained)
7 (autonomy 1M to medication taking constrained)
8 (competence 1M to cessation constrained)
9 (medication taking to cessation constrained)
␹2
df
p
⌬␹2/⌬df
⌬p
IFI
CFI
RMSEA
472.30
537.49
1,010.23
1,018.94
1,019.53
1,046.84
1,022.34
1,022.86
1,022.89
1,022.95
1,044.07
1,030.52
150
150
300
313
314
322
316
318
319
320
321
321
⬍.001
⬍.001
⬍.001
⬍.001
⬍.001
⬍.001
⬍.001
⬍.001
⬍.001
⬍.001
⬍.001
⬍.001
—
—
—
8.72/13
9.31/14
36.62/22
12.12/16
12.64/18
12.67/19
12.72/20
33.85/21
20.30/21
—
—
—
.79
.81
.026
.736
.81
.86
.89
.037
.50
.92
.96
.95
.95
.95
.95
.95
.95
.95
.95
.95
.95
.92
.95
.95
.95
.95
.95
.95
.95
.95
.95
.95
.95
.093
.065
.052
.051
.051
.051
.051
.051
.051
.050
.051
.051
Note. ⌬␹2 represents the difference in chi-square between Model 0 and the specific model in question; ⌬df represents the difference in degrees of freedom
between Model 0 and the specific model in question; ⌬p represents the probability value of the difference in chi-square (given the associated difference
in degrees of freedom) between Model 0 and the specific model in question. IFI ⫽ incremental fit index; CFI ⫽ comparative fit index; RMSEA ⫽
root-mean-square error of approximation; BL ⫽ assessment at baseline; 1M ⫽ assessment at 1 month; AS ⫽ autonomy support.
and thus the components constrained to be equal in that model are
considered equivalent across groups (Byrne, 2001). Results are
presented in Table 5.
Equivalence of the measurement model. Equivalence was
tested by constraining the factor loadings of the indicators on the
latent variables to be equal across the intensive intervention and
the community care conditions (Model 1). Model 1 fit the data
adequately and did not differ significantly from Model 0,
⌬␹2(13) ⫽ 8.72, p ⫽ .79, indicating the latent constructs were
represented and understood equivalently across the two groups.
Equivalence of covariances within the structural model. In
this analysis, the covariance between baseline autonomous motivation and baseline perceived competence was constrained to be
equal (Model 2). Model 2 provided a satisfactory fit to the data and
did not differ significantly from Model 0, ⌬␹2(14) ⫽ 9.31, p ⫽ .81.
This indicates that covariance between latent autonomous motivation and perceived competence at baseline were equal across the
two groups.
Testing the equivalence of the SDT process model. Next, all
structural paths (those between the latent constructs) were constrained to be equal (Model 3). Although Model 3 fit the data
adequately (see Table 5), the fit differed significantly from that of
Model 0, ⌬␹2(22) ⫽ 36.62, p ⬍ .05, indicating that at least one of
the constrained paths in the model is variant across the two groups.
Determination of invariant paths in the SDT process model.
The next three models tested the invariance of the structural paths
between the motivation variables in the SDT process model (Models 4, 5, and 6). Model 4 constrained the paths between baseline
and 1-month autonomous motivation, and between baseline and
1-month perceived competence, to be equal across groups. Model
4 fit the data adequately and did not differ significantly from
Model 0, ⌬␹2(16) ⫽ 12.12, p ⫽ .74. Model 5 constrained the paths
between autonomy support and 1-month autonomous motivation
and between autonomy support and 1-month perceived competence to be equal. The fit of Model 5 was satisfactory and did not
differ significantly from Model 0, ⌬␹2(18) ⫽ 12.64, p ⫽ .81.
Model 6 constrained the path between 1-month autonomous motivation and 1-month perceived competence to be equal. Model 6
provided an adequate fit to the data and did not differ significantly
from Model 0, ⌬␹2(19) ⫽ 12.67, p ⫽ .86. Thus, the paths composing the motivational core of the SDT model were not significantly different between the intensive intervention and the community care groups, and they support our hypotheses that the
process of internalization of autonomy and competence was the
same for patients in both groups.
Three models tested the equivalence across the two groups of
the motivation variables in the SDT process model on medication
taking and abstinence outcomes. Model 7 constrained the path
between 1-month autonomous motivation and medication taking to
be equal across groups. The fit of Model 7 was adequate and did
not differ significantly from Model 0, ⌬␹2(20) ⫽ 12.72, p ⫽ .89,
indicating that regardless of condition, there was a positive relation
between change in autonomous motivation and medication taking.
Model 8 constrained the path between 1-month perceived competence and abstinence to be equal across groups. Model 8 provided
an adequate fit to the data; however, Model 8 did differ significantly from Model 0, ⌬␹2(21) ⫽ 33.85, p ⬍ .05, indicating that the
strength of the path from perceived competence to abstinence was
different in the two groups. Perceived competence at 1 month was
a significant positive predictor of abstinence at 6 months (␤ ⫽ .24,
p ⬍ .001) in the intensive intervention, but competence did not
predict abstinence at 6 months (␤ ⫽ ⫺.05, p ⫽ .50) in community
care. Model 9 constrained the path between medication taking and
abstinence to be equal across the groups. Model 9 provided an
adequate fit to the data, but the fit did not differ significantly from
that of Model 0, ⌬␹2(21) ⫽ 20.30, p ⬍ .50. Thus, these analyses
indicate equivalence of the models except that perceived competence for the intervention patients led to greater abstinence,
whereas it did not for the community care patients. Path coefficients of the models are presented in Figure 2.
Discussion
This clinical trial is the first to demonstrate that an intervention
based on SDT facilitated the internalization of autonomous motivation and perceived competence and that the internalization of
these motivations, in turn, resulted in increased use of cessation
medications and 6-month prolonged abstinence from tobacco. The
MOTIVATING TOBACCO CESSATION
99
Figure 2. Testing the self-determination theory process model across groups for smoking cessation. Model fit:
␹2(332) ⫽ 1,030.52, p ⬍ .001; comparative fit index ⫽ .95; incremental fit index ⫽ .95; root-mean-square error
of approximation ⫽ .051. Double lined paths are invariant across groups. Values represent standardized path
estimates; those in parentheses are for the community care group. *p ⬍ .05. **p ⬍ .01.
current trial also showed that the intervention, which was consistent with the PHS guidelines for intensive treatment of tobacco
dependence, was perceived as being more autonomy supportive
than was community care. The SEM analyses provide evidence
linking the SDT process model to the behavioral outcome, thus
fulfilling a central goal of NIH in issuing the request for applications (Ory, Jordan, & Bazzarre, 2002). Further, mediational analyses were consistent with the causal role the SDT process model
has in tobacco dependence treatment.
In several ways, this study extends the previous empirical support for the process model found in predicting maintenance of
glucose control for diabetes (Williams, McGregor, et al., 2004)
and maintained tobacco abstinence for smokers treated by primary
care physicians (Williams, Gagné, et al., 2002) or through selfhelp (Curry, Wagner, & Grothaus, 1991). This trial is the first to
demonstrate that patient perception of autonomy support can be
significantly increased, as the previous tobacco trial relied on
observer ratings of autonomy support. Second, neither of the
previous interventions significantly increased autonomous motivation or perceived competence compared with the control groups.
Third, this trial confirmed that autonomous motivation and perceived competence represent independent paths to the tobacco
abstinence outcome and that facilitating autonomous use of medications is an important goal of the clinical intervention because it
increased medication taking, which led to greater abstinence.
Multigroup SEM invariance analyses showed that, overall, the
SDT process model was the same in the intensive group as in the
community group. Autonomy support appears to be a critical
factor for enhancing motivation and for promoting behavior
change across the treatment conditions. In particular, paths in the
model from autonomy support to change in autonomous motivation and to change in perceived competence were invariant, as was
the path from change in autonomous motivation to change in
perceived competence. Also, change in autonomous motivation
invariantly predicted use of medications across the two groups.
However, although perceived competence at 1 month predicted
cessation in the intervention group, this relation was not present in
the community care group. Thus, the internalization of autonomy
in the intervention condition led to a change in competence that
independently contributed to abstinence after controlling for medication use, whereas perceived competence internalized from the
community setting did not contribute to abstinence beyond that of
medication use. This difference suggests the following conclusions. First, all hypothesized paths among the SDT variables were
the same in both groups—thus, internalization of autonomy and
competence was confirmed as a proactive process of change in
smoking behavior across types of treatment, as suggested by the
PHS meta-analysis (Fiore et al., 2000, p. 31) and by the U.S.
Preventive Services Task Force (Whitlock, Orleans, Pender, &
Allan, 2002). Second, medication use was the only proximal
predictor of abstinence in community care, whereas competence
and medication use both predicted abstinence in intensive treatment. The intensive treatment was able to provide the kinds of
interpersonal supports that bolstered perceived competence and
promoted abstinence independent of medication use. The failure of
community care competence to independently predict cessation is
consistent with community practice. The community care physicians were less autonomy supportive, and the result was that their
patients were less autonomously motivated to use the medications.
Further, the community physicians were not likely to have spent
100
WILLIAMS ET AL.
much time working to build a quit plan, nor on actually engaging
their patients in the problem solving and skills building needed for
perceived competence and sustained change. SDT uniquely predicts and demonstrates that only when perceived competence is
accompanied by the experience of autonomy will it motivate
sustained change.
We note that, in the intention-to-treat analyses, autonomous
motivation for cessation was increased from baseline to 1 month
by the intervention only for a subgroup of patients who chose to
see both a counselor and a study physician. The failure to increase
autonomous motivation for cessation for all smokers in the intervention group may have occurred because the 1-month follow-up
was measured before the full intervention was experienced and
because 140 participants did not return after the first visit, thus
limiting the effect of the intervention. The fact that autonomous
motivation for cessation was increased in the subgroup of patients
who met with both a study physician and a counselor is consistent
with the PHS meta-analysis and 5A’s (ask, advise, assess, assist,
and arrange) model, which indicate a unique effect for physicians
in counseling smokers to quit (Fiore et al., 2000; Whitlock et al.,
2002), and suggests that physician counseling may have its effect
by increasing autonomous motivation for cessation.
This randomized trial also extends earlier findings from observational data (Williams, Rodin, et al., 1998) that providing a more
autonomy-supportive intervention increases autonomous motivation for taking medications and self-reported medication taking.
The standardized parameter estimates of .2 or greater, and NNTs
of near 10, reflect the intervention being clinically significant
(Woolf, 1999). A recent meta-analysis of communication in medicine (Epstein, Alper, & Quill, 2004) found a paucity of evidence
on how communication style can improve patient participatory
decision making and highlights the importance of the empirical
evidence that this study provides. Using SDT to guide clinicians in
motivating patients to take medications seems likely to have an
important effect in tobacco dependence treatment (Pierce &
Gilpin, 2002) and in adherence to medications in general.
The 6-month nonvalidated point prevalence in the current study
was 14.1% versus 5.8% (OR ⫽ 2.7). This absolute quit rate and
odds ratio compare favorably to the PHS meta-analysis 5-month
nonvalidated point prevalence of 16.8% versus 10.8% and OR ⫽
1.7 for 4- to 8-session intensive treatments (Fiore et al., 2000,
Table 17, p. 63). It is possible that the slightly lower absolute
abstinence rate found in this intervention (14.1% compared with
16.8% in the PHS meta-analysis) and moderately lower community control absolute abstinence rate in this report (5.8% compared
with 10.8% in the PHS meta-analysis) may be due in part to the
inclusion of patients who did not initially intend to quit (52%),
although there was no significant correlation between intention to
quit and the outcome in either group. Another possible explanation
of the lower abstinence rates is that we accepted participants with
depression, anxiety, and chemical dependence disorders, who are
believed to have lower quit rates than populations typically accepted into tobacco dependence trials. Finally, our patients were
poor and undereducated relative to the averages in their county,
and these factors may have contributed to the lower quit rates, for
example, by making use of medications more difficult.
Intensively treated patients’ self-reported use of medication and
abstinence from tobacco were enhanced whether or not they initially intended to quit. These results add to a growing body of
literature showing that patients who do not indicate an initial
willingness to stop will stop if they receive appropriate interventions (Curry, McBride, Grothaus, Louie, & Wagner, 1995; Carpenter, Hughes, Solomon, & Callas, 2004). Together, findings
from these studies support the PHS guidelines’ (Fiore et al., 2000)
recommendations for use of the 5R’s (relevance, risks, rewards,
roadblocks, and repetition) for those who don’t want to quit.
Because the intervention was wholly consistent with the PHS
guidelines, the results provide the first direct empirical support for
autonomy, competence, and autonomy support being three of the
psychological variables through which the PHS-recommended intensive treatment works. The fact that medication taking strongly
predicted abstinence in both groups, in combination with Carpenter et al.’s (2004) findings that offering medications before smokers stop helps them quit, suggests that the autonomy-supportive
discussion of use of cessation medications may be beneficial in all
tobacco-dependence interventions, whether or not smokers initially indicate they want to quit.
Limitations of the study include that there was a larger percentage of non-Whites than Whites who dropped out of the study.
Continued effort is needed to understand how to better retain
minority patients, perhaps by including minority counselors and
physicians. Also, patients who felt more autonomous and competent at the start of the study were more likely to drop out. This may
be consistent with SDT, as these patients with higher levels of
motivation may have felt better able to quit smoking on their own,
thus feeling less need to stay in the program. Also, mediators from
other theories were not tested, leaving the possibility that other
mediators measuring social support, stages of change, or selfefficacy should be tested in future behavior change trials (Fiore et
al., 2000; Whitlock et al., 2002). Nonetheless, initial support for
the SDT variables mentioned in the PHS guidelines has been
found. An additional limitation comes from our accepting only a
portion of the population with mental illness, which is ravaged by
tobacco-related disease. These limitations indicate additional research is called for that includes patients with other mental illnesses and that is conducted in other intensive treatment centers
that have obtained higher quit rates.
In conclusion, an intensive individual tobacco treatment intervention that focused on supporting patients to make a clear choice
about whether or not they want to smoke was more effective in
promoting abstinence than was community care. The intervention
was found to have its effects by facilitating the internalization of
autonomous and competence motivations. The utility of providing
interventions for tobacco use to those who do not intend to quit
was also supported. These data support that the PHS Clinical
Practice Guideline Treating Tobacco Use and Dependence (Fiore
et al., 2000) is effective in part by supporting autonomy (intratreatment support), which results in greater volition (autonomy for
cessation and for using medications), which leads to greater competence (through skills building and problem solving) and use of
medications to promote abstinence from tobacco.
References
Arbuckle, J. L. (2003). AMOS (Version 5) [Computer software]. Chicago:
SmallWaters.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York:
Freeman.
MOTIVATING TOBACCO CESSATION
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
51, 1173–1182.
Bollen, K. (1989). Structural equations with latent variables. New York:
Wiley.
Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic
concepts, applications, and programming. Mahwah, NJ: Erlbaum.
Carpenter, M. J., Hughes, J. R., Solomon, L. J., & Callas, P. W. (2004).
Both smoking reduction with nicotine replacement therapy and motivational advice increase future cessation among smokers unmotivated to
quit. Journal of Consulting and Clinical Psychology, 72, 371–381.
Curry, S. J., McBride, C., Grothaus, L. C., Louie, D., & Wagner, E. H.
(1995). A randomized trial of self-help materials, personalized feedback,
and telephone counseling with nonvolunteer smokers. Journal of Consulting and Clinical Psychology, 63, 1005–1014.
Curry, S., Wagner, E. H., & Grothaus, L. C. (1991). Evaluation of intrinsic
and extrinsic motivation interventions with a self-help cessation program. Journal of Consulting and Clinical Psychology, 59, 318 –324.
Dale, L. C., Glover, E. D., Sachs, D. P., Schroeder, D. R., Offord, K. P.,
Croghan, I. T., et al. (2001). Bupropion for smoking cessation: Predictors of successful outcome. Chest, 119, 1357–1364.
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and selfdetermination in human behavior. New York: Plenum Press.
Epstein, R. M., Alper, B. S., & Quill, T. E. (2004). Communicating
evidence for participatory decision making. Journal of the American
Medical Association, 291, 2359 –2366.
Fagerstrom, K.-O., & Schneider, N. G. (1989). Measuring nicotine dependence: A review of the Fagerstrom Tolerance Questionnaire. Journal of
Behavioral Medicine, 12, 159 –182.
Farkas, A. J., Pierce, J. P., Zhu, S. H., Rosbrook, B., Gilpin, E. A., Berry,
C., et al. (1996). Addiction versus stages of change models in predicting
smoking cessation. Addiction, 91, 1271–1280.
Fiore, M. C., Bailey, W. C., Cohen, S. J., Dorfman, S. F., Goldstein, M. G.,
Gritz, E. R., et al. (2000). Treating tobacco use and dependence.
Rockville, MD: U.S. Department of Health and Human Services, Public
Health Service.
Grundy, S. M., Pasternak, R., Greenland, P., Smith, S., Jr., & Fuster, V.
(1999). Assessment of cardiovascular risk by use of multiple-risk-factor
assessment equations: A statement for healthcare professionals from the
American Heart Association and the American College of Cardiology.
Circulation, 100, 1481–1492.
Hughes, J. R., Keely, J. P., Niaura, R. A., Ossip-Klein, D. J., Richmond,
R. L., & Swan, G. E. (2003). Measures of abstinence in clinical trials:
Issues and recommendations. Nicotine and Tobacco Research, 5, 13–25.
Kline, R. B. (1998). Principles and practice of structural equations modeling. New York: Guilford Press.
Murray, R. P., Connett, J. E., Rand, C. S., Pan, W., & Anthonisen, N. R.
(2002). Persistence of the effect of the Lung Health Study (LHS)
smoking intervention over eleven years. Preventive Medicine, 35, 314 –
319.
National Heart, Lung, and Blood Institute. (1997). Cholesterol lowering in
the patient with coronary heart disease: Physician monograph. (NIH
Publication No. 97-3794.) Bethesda, MD: National Heart, Lung, and
Blood Institute.
National Institutes of Health (National Heart, Lung, and Blood Institute).
(1997). Cholesterol lowering in the patient with coronary heart disease.
[Physician monograph. NIH Publication No. 97-3794] Bethesda, MD:
Author.
101
Ory, M. G., Jordan, P. J., & Bazzarre, T. (2002). The Behavior Change
Consortium: Setting the stage for a new century of health behaviorchange research. Health Education Research, 17, 500 –511.
Pierce, J. P., & Gilpin, E. (2002). Impact of over-the-counter sales on
effectiveness of pharmaceutical aids for smoking cessation. Journal of
the American Medical Association, 288, 1260 –1264.
Pierce, J. P., & Gilpin, E. (2003). A minimum 6-month prolonged abstinence should be required for evaluating smoking cessation trials. Nicotine and Tobacco Research, 5, 151–153.
Pojer, R., Whitfield, J. B., Poulos, V., Eckhard, I. F., Richmond, R., &
Hensley, W. J. (1984). Carboxyhemoglobin, cotinine, and thiocyanate
assay compared for distinguishing smokers from non-smokers. Clinical
Chemistry, 30, 1377–1380.
Prochaska, J. O., DiClemente, C. C., & Norcross, J. C. (1992). In search of
how people change: Applications to addictive behaviors. American
Psychologist, 47, 1102–1114.
Public Health Service, U. S. Department of Health and Human Services.
(2000). You can quit smoking. Consumer guide (ISSN-1530-6402).
Ryan, R. M., & Connell, J. P. (1989). Perceived locus of causality and
internalization: Examining reasons for acting in two domains. Journal of
Personality and Social Psychology, 57, 749 –761.
Sheldon, K. M., Williams, G. C., & Joiner, T. (2003). Self-determination
theory in the clinic: Motivating physical and mental health. New Haven,
CT: Yale University Press.
U.S. Department of Health and Human Services. (1990). The benefits of
smoking cessation: A report from the Surgeon General (DHHS Publication No. CDC 90 – 8416). Atlanta, GA: U.S. Department of Health and
Human Services, Public Health Service, Centers for Disease Control,
Center for Chronic Disease Prevention and Health Promotion, Office on
Smoking and Health.
Whitlock, E. P., Orleans, C. T., Pender, N., & Allan, J. (2002). Evaluating
primary care behavioral counseling interventions: An evidence-based
approach. American Journal of Preventive Medicine, 22, 267–284.
Williams, G. C., Cox, E. M., Kouides, R., & Deci, E. L. (1999). Presenting
the facts about smoking to adolescents: Effects of an autonomysupportive style. Archives of Pediatrics and Adolescent Medicine, 153,
959 –964.
Williams, G. C., Freedman, Z. R., & Deci, E. L. (1998). Supporting
autonomy to motivate patients with diabetes for glucose control. Diabetes Care, 21, 1644 –1651.
Williams, G. C., Gagné, M., Ryan, R. M., & Deci, E. L. (2002). Facilitating
autonomous motivation for smoking cessation. Health Psychology, 21,
40 –50.
Williams, G. C., Grow, V. M., Freedman, Z. R., Ryan, R. M., & Deci, E. L.
(1996). Motivational predictors of weight loss and weight-loss maintenance. Journal of Personality and Social Psychology, 70, 115–126.
Williams, G. C., McGregor, H. A., Zeldman, A., Freedman, Z. R., & Deci,
E. L. (2004). Testing a self-determination theory process model for
promoting glycemic control through diabetes self-management. Health
Psychology, 23, 58 – 66.
Williams, G. C., Minicucci, D. S., Kouides, R., Levesque, C. S., Chirkov,
V. I., Ryan, R. M., & Deci, E. L. (2002). Self-determination, smoking,
diet and health. Health Education Research, 17, 512–521.
Williams, G. C., Rodin, G. C., Ryan, R. M., Grolnick, W. S., & Deci, E. L.
(1998). Autonomous regulation and long-term medication adherence in
adult outpatients. Health Psychology, 17, 269 –276.
Woolf, S. H. (1999). The need for perspective in evidence-based medicine.
Journal of the American Medical Association, 282, 2358 –2365.